SOT: In phase noise reduction audio question

Robert W wrote on 2/28/2009, 7:04 AM

I have a piece of mono cassette tape which I have captured with a stereo playback mechanism. I am sure I used to have a trick that was essentially the reverse of OOPS (Out Of Phase Stereo). Instead of rejecting everything in the signal that was present on both the left and right track, it rejected all the differences between the left and right track. As tape noise is not consistent across the width of the cassette, this was a handy noise reduction trick when working with stereo transfers of mono tapes.

However, I can not remember how I did it. I am pretty sure I did it for someone with Sound Forge 3 or 4, and I used no plugins, just the basic editing tools. Can anybody show me a method to preserve similar audio in two stereo sound, while rejecting the differences?

Comments

blink3times wrote on 2/28/2009, 7:55 AM

I'm not 100% sure what your end result wants to be but by the sound of it you simply want to either isolate or remove the difference between 2 tracks. If I have that right then simply invert one track then mix. This will leave you with the difference between the 2 tracks. If you wish to REMOVE the difference then just take the difference mix that you just produced and mix that with each of the original tracks.

Robert W wrote on 2/28/2009, 8:06 AM

Thanks for the attempt, but unfortunately that approach will not work. Half of the noise will be out of phase, so when you add it back to the mix that part of the noise will be louder.

I think it may be too fiendish to work out what I did. However, an associate has Adobe Audition, and thinks that can do it, so I am going to visit him and see what happens.

blink3times wrote on 2/28/2009, 9:04 AM

Audition is a MUCH better bet. There are some great filters in there.

farss wrote on 2/28/2009, 2:47 PM

What Blink described will do exactly what you asked for and as you noted not achieve much as the noise is indeed random and therefore has no phase correlation. You simply cannot cancel noise that way.

What you can do is to Add the two tracks. This will work because the wanted signal is in phase on both tracks so it always adds. The noise being random does not add in the same way. This is drop dead easy to achieve in Vegas. Under clip properties select Channels>Combine. Job done.

This will only work well if the tape and player is in good condition. Stretch in the tape, missaligned heads or guides etc both in the player or the original recorder will cause the wanted signal to not be precisely in phase on both channels so when you add both channels you'll introduce a nice comb filter into your audio, yucky.

I've done a LOT of archival audio work. Even master tapes recorded at 15ips can have have problems trying to combine the two tracks into mono. I generally would select the cleanest channel/track and use Vegas to duplicate that accross both stereo channels. At 15ips tapes were pretty clean so tape noise wasn't worth the effort trying to reduce.

It's possible that the original cassette was recorded with one of the Dolby NR techniques. Check this when you play the tape as the ONLY way to correctly decode the signal is with the hardware, Dolby A, B and C has never been implemented in software....that I can find.

Bob.

Robert W wrote on 3/1/2009, 3:09 AM

Thanks for the tips Bob. I tried collapsing to mono, thinking as you suggested that the chances are some noise will be canceled out by luck, and it did yield a noticeably improved signal to noise ratio. However, the signal level was still very poor compared to the noise level, so I borrowed my friends setup to try a slightly different approach. First I used the Raygun Pro Direct X plugin to remove rumble and hum. Then I used the Adobe Audition Stereo Image/Center Channel Extractor to analyze and separate the clean part of the signal from the noise. I swiftly discovered that this can cause noise reduction type problems if not used carefully, but I reached a good compromise, and on a much better signal to artifacts ratio than I was getting with my noise reduction tests.